tag:blogger.com,1999:blog-31545676.post2525046768828437357..comments2020-01-16T18:27:47.773-05:00Comments on Sabermetric Research: How Elo ratings overweight recent resultsPhil Birnbaumhttp://www.blogger.com/profile/03800617749001032996noreply@blogger.comBlogger7125tag:blogger.com,1999:blog-31545676.post-75555745907393137072018-11-30T14:42:16.594-05:002018-11-30T14:42:16.594-05:00This all makes sense. I definitely recall making a...This all makes sense. I definitely recall making a logistic regression model using data from the last X-number of MLB games (plus some other non temporal variables). It was fairly basic, and compared it to the ELO probabilities, it was a superior predictor every year except 2014. Pkdryan2https://www.blogger.com/profile/12023342548271502075noreply@blogger.comtag:blogger.com,1999:blog-31545676.post-43967963758994547662018-09-10T19:41:50.573-04:002018-09-10T19:41:50.573-04:00Sorry so late getting to this ... I forgot.
FiveT...Sorry so late getting to this ... I forgot.<br /><br />FiveThirtyEight used 1500 for a .500 team. If you use the formula from the wikipedia page, you'll see that a 10-point difference from .500 works out to .514. <br /><br />Specifically, 1 divided by (1 + 10^(10/400)) equals .514. In that formula, 10 is the point difference (1510-1500), and 400 is the Elo constant.Phil Birnbaumhttps://www.blogger.com/profile/03800617749001032996noreply@blogger.comtag:blogger.com,1999:blog-31545676.post-80324323462569139162018-09-01T16:43:07.707-04:002018-09-01T16:43:07.707-04:00How did you get this " ...1510 represents a ....How did you get this " ...1510 represents a .514 record ". I think your math is completely off there.<br /><br />I might be wrong, so I wanted to double check.<br /><br />Thanks in advance.Unknownhttps://www.blogger.com/profile/03272402840602267890noreply@blogger.comtag:blogger.com,1999:blog-31545676.post-40962721218484616722017-11-19T05:32:45.829-05:002017-11-19T05:32:45.829-05:00Really good post, as usual. One comment:
"Wh...Really good post, as usual. One comment:<br /><br />"When you have a large, visible shock to team talent, I don't see why you wouldn't just adjust for it based on fundamentals, instead of waiting a whole season for your formula to figure it out."<br /><br />Because you are not always right about its true impact, and learning it incrementally is the preferred strategy averaged over all cases. An injured star may be replaced by a teammate who is actually surprisingly better, for instance.<br />I agree with the post though. Bayesian methods dominate Elo, but they are more difficult to understand for all consumers of the ratings output. Longitudinal changes to Elo are easier to follow than modifications to probability distributions.James Willoughbyhttps://www.blogger.com/profile/14205227458245910126noreply@blogger.comtag:blogger.com,1999:blog-31545676.post-73187350136550080392017-11-17T17:49:45.154-05:002017-11-17T17:49:45.154-05:00To make the last game worth twice as much as the f...To make the last game worth twice as much as the first instead of 4 times as much, the K factor would need to be reduced. However this will cause the initial rating to have a higher value. I believe the best way to improve Elo ratings (though this will still have problems) is that to have a flexible K based off of how many games have been played and the number of time since the last game.<br /><br />Honestly my biggest problems with ELO ratings are the SOS adjustments. If you beat a 1900 team with true talent of 1500 than your rating will shoot up a ton even though the other teams rating will most likely drop. ELO's SOS is entirely based on the games that happen before the date and ignores anything that happens in the future even though it gives valuable information on their true talent during the game.Unknownhttps://www.blogger.com/profile/12476950425455746927noreply@blogger.comtag:blogger.com,1999:blog-31545676.post-22544999415563023622017-11-17T16:49:27.446-05:002017-11-17T16:49:27.446-05:00I'm not sure I understand the question. Elo m...I'm not sure I understand the question. Elo might predict, at the beginning of the year, that the team has a .500 talent. But then that estimate changes every game.Phil Birnbaumhttps://www.blogger.com/profile/03800617749001032996noreply@blogger.comtag:blogger.com,1999:blog-31545676.post-26659086805401290142017-11-17T15:29:48.873-05:002017-11-17T15:29:48.873-05:00If ELO predicts a team will have a .500 record and...If ELO predicts a team will have a .500 record and at the end of the season they do, should it matter that the reason they do is they lost their star player for half the season? What's the talent estimation of injury luck?Zachhttps://www.blogger.com/profile/01115366572915518720noreply@blogger.com